An article written by Drug Store News' Nigel F. Maynard. Find the original article here.
A panel of industry executives discussed the accelerated growth of artificial intelligence and how one subset, Generative AI, will affect the way retail and healthcare business function in the near future.
Generative artificial intelligence generates text, images, or other media, using generative models. This type of AI models learn from their input training data and then generate new data with characteristics that are the same or similar in certain ways.
To set the tone for the day, moderator Deborah Weinswig, CEO and founder of Coresight Research, talked about the rapid rise of Generative AI and its impact on the business community: one study suggested that Generative AI will have a $9.2 Trillion impact on retail by 2029.
This panel looked at ways AI could create new content and ideas, including conversations, stories, images, videos, and music that will lead to increased sales, better margins and lower costs.
Panelists included Andre Persaud, Consultant, who most recently was an executive at Rite Aid; Lori Schafer, CEO, Digital Wave Technology; and Guy Yehiav, president, SmartSense by Digi.
Weinswig says Generative Ai is a powerful tool but warned that users should be mindful of the technology and its “hallucinations,” which is when AI makes up false information or facts that aren't based on real data or events. “This is the power of generative AI,” she said. “It fills in, and it may be false often when it fills in.”
The panelists gave their thoughts on the topic of generative AI in healthcare.Image
Schafer:
“We always research the latest technology. And when we saw the power of what generative AI could do … we saw that and we said, ‘this can literally revolutionize the way businesses work and we have to be part of it.’ And so we started last year before chat GPT was ever well known. And I will say this year has been the most wild ride. I've been in the business for twenty something years on the technology side, and I've never seen anything like it because it is able to really improve efficiencies within the business, and at the same time greatly improve customer experience, greatly improve your ability to generate more conversions regardless of whatever industry you're in.”
“We're at the beginning of this, don't be afraid of it. Use it as a co-pilot. It's not magic. It's actually not even intelligent, even though it's called generative artificial intelligence. I know how these algorithms work in the background and from a standpoint, you know, get a partner to work with that can listen to your problem and create the use case quickly and then learn from that. So that's my advice.”
Yehiav:
“At SmartSense by Digi “we help healthcare and pharmacy and CPG manufacturing, groceries and restaurants optimize their labor. So it is making sure that the products have the right quality while the people that work, that serve the clients and not think about all the regulations and the compliance that continue to change every day. It speaks specifically in pharmaceutical. You have that, uh, continuous changes, uh, in, in food as you know, uh, FSMA 2026 is coming with traceability that is similar to what came out last year for pharmaceutical. So, you know, one vertical is learning from the others, and sensors have been here for a while, but doing something with the data, optimizing it and telling people what they should do in order to optimize the outcome, that's what we do.”
“Yeah, so, you know, condition monitoring has been here for a while. I think the most critical is reducing those false positives. If I tell you every five minutes, ‘Hey, you left the door open’ and you look and you say, ‘Hey, the door is closed.’ Next time I'll tell you that we will not look at it. So reducing false positives is where generative AI can help because it looks at all the historical data then tells you “Hey, you should close the door’ and you can penalize or credit the model based on if you see the door close or open. And then there's a feedback loop that continues to learn and then use different words. So it's not just prescriptive action, but it's about that specific store that you are in with specific assets. It learns about the asset.”
“I think that to start, you look at the specific cases that you're trying to improve and then it's easier to build the data model for those. So you look at the specific sensors, you look at the telemetry data that you need, and you run the model. And typically when you run just a model, you'll be 50% wrong. Now you say, “Hey, humans, make less mistakes. But, that's the beginning. That's the first four or five weeks. What you then do is you start penalizing the model and credit the model based on the trueness of the result. And then when it reaches about 90%, you can now start releasing it. …And so starting those models is important to have a specific case that you're trying to capture, take the data, run the model, learn, and then start using it to improve.”
Persaud:
“AI allows a much easier, much more predictable process in demand planning. And what I would share also from a practical perspective is that I've worked for three organizations who have implemented the same AI enabled demand planning system, only one of the three worked. And the reason why is the other two just copy and paste everything they were doing on Excel and put it into the demand planning rather than allow the system to work. So, part of this is trust.”
“I think the acceleration is allowing the process to work at more mid-management levels versus having to get C-suite approval for everything. I think C-Suite should be approving the dollars. C suite should be approving the governance and the framework. Well, that's a team role, right? And, you know, just going back to the demand planning piece, I allowed our supply chain team to run with that. I did not want to get involved. And it worked there. The other organizations where it hasn't worked, everybody had their finger on it, including the CFO. So you come back to this theme of trust. The other thing is these systems are very iterative. I flip to marketing. How many times have you had a marketing department or chief marketing officer come and say, no, we need to improve our, our top line. I can run these six campaigns and based upon cutting the customers in five or six different ways, you know, the reality is with an AI driven process, there's no more of that. The AI system has every customer attribute and understands how they shop. And you could be running hundreds of campaigns simultaneously and, and very easily learn through that process over a period, over a day, two days, three days, which ones are the most effective. So it does speed things up significantly. There was a retailer that employed that approach and saw over a 10 x improvement on conversion rate.